FedPCD: Priority-Constrained Federated Aggregation

Key points

• Federated learning (FL) seeks to optimize a shared global model from decentralized client data.1
• Existing fairness-aware and multi-objective FL methods address this through client re-weighting, gradient correction, or common-descent formulations, but often change the global optimization direction or treat client objectives symmetrically.1
• FedPCD preserves the conventional federated aggregate as the primary objective, while allowing only the minimum correction necessary to meet client performance constraints.1
• FedPCD further preserves competitive trade-offs between global accuracy and client performance variability.1

Sources (1)

• [1]FedPCD: Priority-Constrained Federated Aggregation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 12:34 PM

  We propose FedPCD, a priority-constrained federated optimization framework that separates the global optimization objective from client protection.

  Federated learning (FL) seeks to optimize a shared global model from decentralized client data.

Extractive summary: sentences quoted from the sources.